All decisions based on inferential statistics are based on 4 criteria:
- Normality of the scores
- Type of sample group
- Number of independent variables
- Number of dependent variables
If variables are normally distributed, parametric forms of statistics are used.
If variables are not normally distributed, non-parametric forms of statistics are used.
- For correlation:
- Normally distribution use Pearson’s Correlation
- Not normal distribution use Spearman’s correlation
- For group comparisons:
- Both parametric and non-parametric tests are dependent on the type of group
- If sample groups which are to be compared are independent of each other then different statistic is used, and
- If the sample group is the same, or related, such as that for pre-post intervention design, then a different group of statistic is used.
• Therefore, before deciding on a test, the researcher must look for three things:
- Normality of the variable
- If the sample groups are independent or related
- Number of variables present